On Fri, Aug 03, 2018 at 01:03:36AM +0000, Bart Van Assche wrote:
> On Fri, 2018-08-03 at 08:11 +0800, Ming Lei wrote:
> > On Fri, Aug 03, 2018 at 12:08:54AM +0000, Bart Van Assche wrote:
> > > On Fri, 2018-08-03 at 07:53 +0800, Ming Lei wrote:
> > > > blk_pm_add_request() calls pm_request_resume() for waking up device, but
> > > > it is wrong because it is async request, which can't guarantee device
> > > > will be ready before calling .queue_rq().
> > >
> > > That's a longstanding issue and not something that has been introduced by this
> > > patch series. Hence patch 6/9 that issues a warning if pm_request_resume() would
> > > be called.
> >
> > It is introduced by your patch, please see blk_pm_allow_request().
>
> Did you perhaps misread my patch? Please have a look at commit c8158819d506
> ("block: implement runtime pm strategy"). I think that commit from 2013
> introduced the pm_request_resume() call in blk_pm_add_request().
>
> By the way, if you would like to see that call removed, I'm fine with
> removing that call.
pm_request_resume() isn't used wrong now, and the issue is introduced by
removing blk_pm_allow_request(), which is called in elv_next_request() for
avoiding this issue, so that requests can be kept in scheduler queue if
device isn't active.
Thanks,
Ming
